Repeatability estimates of hybrid diploids (AA) of banana

Abstract

The objective of this work was to estimate the coefficients of repeatability in diploid banana hybrids, and predict the number of required measurements for main quantitative traits. The evaluated hybrids were: 042079‑06, TH03‑01, 089087‑01, 003023‑03, 013018‑01, 001016‑01, 086094‑20, 013004‑06, and 091079‑03. Twelve agronomic traits were measured in three cycles. For the repeatability coefficient estimate, the following analyses were performed: analysis of variance, principal components from the matrix of phenotypic variance and covariance analysis, and the correlation and structural analyses. For each trait, the minimum number of measurements to predict the actual value of the hybrids was determined. Genetic variability was observed among genotypes. Repeatability estimates were high, showing the regularity of accessions. One to five measurements are needed to predict the actual value of production characters. Methods based on principal component analysis are more efficient to estimate the coefficient of repeatability.